Emotions Recognition and Signal Classification: A State-of-the-Art

Introduction

Affective Computing (AC) is specializes in creating of and interacting with systems that are able to detect and respond to affective states (Picard, 2002). The AC field is based on the contributions of numerous domains, such as mathematics, engineering, education, computer science, sociology, physiology, and psychology. These several contributions reflect to which extent it is very complex to describe, understand and finally emulate the feeling experience. In fact, the term AC is quite new; it was first addressed in late the 1990s by Rosalind Picard (Picard, 1997). The main aim of AC is to establish more comfortable interaction between the human and computers (Hore, 2019); since the absence of emotions makes the humans feel uncomfortable and annoyed (Beale, 2008). Another opinion proposed by Clifford Nass, who revealed that humans deal with computers as they deal with each other (Nass, 1997). Hence, the emotionally respond of computers to the users is a must. AC show great benefits in different aspects such as in learning, the e-learning gains a severe attention and become favored by many students and enhancing the communication between students and computer improves the e-learning experience and the student mental state (Picard, 2001), and in social context (Breazeal, 2003).

The primary step to achieve an effective affective computing experience is detecting or recognizing emotions. Enabling a computer or any device to recognize emotions needs software and hardware requirements, examples of the hardware requirements are physiological sensors, video and depth sensors, and the standard input devices. The physiological sensors are used to measure the blood volume pulse, temperature, the skin conductance, and respiratory signal. Nevertheless, physiological sensors have a disadvantage that they are uncomfortable for users as well as these sensors are not practical and impossible to be used in the real life situations which require detecting emotions during daily working and learning processes, so they are usually used to verify the accuracy of a specific classifier. Video sensors like cameras grant information about the facial expression (Dey, 2019). The complexity of this method is due to the need of complex pattern recognition and image processing, whereas the depth sensors are often based on infrared light technology and used to detect human movements and gestures. The standard input devices such as the conventional mouse and keyboard represent an effective and unobtrusive method to collect the data from users; its effectiveness is for the no need of additional hardware to be added to the computer (Kołakowska, 2014). Computer or devices with such equipment result in affect-sensing systems, which are used in the field of affective computing to refer to systems that are able to detect emotions using information extracted from received patterns and signals (Daily, 2017).